Friday 17 April 2009

Exchange 2010 Hub Transport feature, Shadow Messaging/Redundancy

Fed up with RAID10 for your Hub Transport Servers? Well half the cost of your Hub Transport disk subsystem with Exchange 2010.

Exchange 2010 Hub Transport Servers support a feature called Shadow Redundancy. This means that the HT Server gets confirmation the other HT it passed it onto actually successfully passed it on itself.

Great you say! But how does this benefit me? Well for a start your users won't have to resend lost messages, and well secondly you don't necessarily have to rely on RAID10 anymore. RAID0 is a real possilbility for your Hub Transport Servers. You still need your IO (when you are transferring a lot of messages and have features like Journaling or Transport Rules are setup and enabled), but you now have the potential to half your disk cost.

Take an Enterprise that has 20 Hub Transport Servers, they may have a 6 disk SAS based RAID10 solution for their mail queue database. That's 120 SAS disks. With Shadow Redundancy you can half that to 60, with just as good or slighter better IO (no write penalties).

If a disk fails then you just rip it out and rebuild the RAID set, and then reboot your Server or restart your Microsoft Exchange Transport Service - you don't have to worry about the lost mail in that mail database as previous hops will have not recieved confirmation of delivery and they will re-route using alternative HT Servers.

Now RAID0 isn't great in all circumstances. You wouldn't use RAID0 for any Hub Transport Servers that recieved mail directly from the Internet for example. However this is a perfect solution for Hub Transport Servers in dedicated Hub Transport AD Sites - where they only ever pass mail onto Hub Transport Servers in other AD Sites.

Note this feature isn't support with communication between Exchange 2010 and 2007 Hub Transport Servers, It's 2010 to 2010 only.

Oliver Moazzezi

MVP - Exchange Server

No comments: